Laser wire feeding process parameter control method and system based on machine learning

Abstract

The invention provides a laser wire feeding process parameter control method and system based on machine learning, and belongs to the field of laser wire feeding. The problems that in an existing laser wire feeding process, process parameters need to be manually corrected, efficiency is low, and manual correction errors are relatively large are solved. According to the method, the bead height and the bead width of the monorail sediment are obtained by testing different laser powers, scanning speeds and wire feeding speeds, then a fitting model is obtained from a data set by using a BP neural network, and the model is used for technological processing. By utilizing the process parameter data sample, the bead height and the bead width of the sediment can be well predicted through the process parameters, and the required process parameters can be reversely predicted through the required bead width and the bead height, so that the shape deviation of a laminated object caused by wrong judgment of observing a molten pool in real time by a technician can be reduced, and the quality of the product is improved. And equipment data can be mined and analyzed, the equipment is endowed with the capability of intelligently processing the technological process, and the professional requirement of the technological process on technologists is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of laser wire feeding, and in particular to a method and system for controlling laser wire feeding process parameters based on machine learning. Background Art

[0002] Laser direct metal deposition (LDMD) is an additive manufacturing method that can provide metal cladding layers with good corrosion resistance, good surface finish and precise geometry on substrates, and has been widely used in aerospace, energy and power, metallurgical machinery and other fields. However, the process stability of laser wire cladding is one of the important factors limiting its industrial application. Therefore, the control of laser wire feeding process parameters is of great significance.

[0003] In direct energy deposition, the laser and wire metal interact, liquefy and deposit on the substrate material. Due to the complexity of the entire deposition process, it is difficult to directly construct a very accurate physical model that can accurately show the relationship between process parameters and the width and height of the deposited beads.

[0004] The traditional control method is to preset process parameters based on experience, and then manually adjust the processing parameters by observing the state of the deposits and the molten pool. However, this process of correcting process parameters through manual experience is often not the most reasonable, and the processing requires technical technicians to observe the entire process, which is inefficient. Summary of the invention

[0005] The technical problems to be solved by the present invention are:

[0006] In order to solve the problem that the existing laser wire feeding process requires manual correction of process parameters, which is inefficient and has relatively large manual correction errors.

[0007] The present invention adopts the following technical solutions to solve the above technical problems:

[0008] The present invention provides a method for controlling laser wire feeding process parameters based on machine learning, comprising the following steps:

[0009] S100, on the same test plate, respectively changing the process parameters of the equipment to obtain different single-track deposits, measuring the bead height and bead width of the single-track deposits to obtain an initial data set, wherein the process parameters include laser power, scanning speed, and wire feeding speed;

[0010] S200, dividing the initial data set collected in step S100 into a training set and a test set according to a certain ratio, training the training set using a BP neural network to obtain a training model, and then using the test set to detect the fitting degree of the training model to obtain the best fitting model;

[0011] S300, batch importing process parameters into the best fitting model obtained in step S200, obtaining the corresponding bead height and bead width data of the sediment, acquiring the database, and calling the process parameters in the database for processing according to the required layered shape.

[0012] Further, in step S200, the laser power, scanning speed and wire feeding speed are set as input parameters X of the BP neural network, and the bead width and bead height are set as output Y of the neural network;

[0013] Given training samples S = {(X1, Y1), (X2, Y2), ..., (X n , Y n )},X i ∈R 3 , Y i ∈R 2 , X is the normalized process parameter data, R 3 is the dimension of the input data, i.e., laser power x1, scanning speed x2, wire feeding speed x3;

[0014] Set the threshold of the jth neuron in the output layer with σ j Indicates that the threshold of the mth neuron in the hidden layer is τ m Indicates that the connection weight between the i-th neuron in the input layer and the m-th neuron in the hidden layer is λ im , the connection weight between the mth neuron in the hidden layer and the jth neuron in the output layer is μ mj ; The input received by the mth neuron in the hidden layer is:

[0015]

[0016] The input received by the jth neuron in the output layer is:

[0017]

[0018] Among them, h m represents the value of the mth neuron in the hidden layer, and q represents the total number of q neurons in the hidden layer.

[0019] Furthermore, the discriminant function between the hidden layer and the output layer uses the Sigmoid function For the training set, assume that the current neural network output is Then the mean square error of the BP neural network on the data set is:

[0020]

[0021] Furthermore, the gradient descent algorithm is used to calculate the target E k The negative gradient direction is used to adjust the parameters. k, given the learning rate ξ, after one round of learning:

[0022] Δμ hj =ξf j h m

[0023] Δσ j =-ξf j

[0024] Δλ ih =ξl h x i

[0025] Δτ h =-ξl h

[0026] in:

[0027] The learning rate ξ controls the step size of each learning iteration.

[0028] Furthermore, the value of ξ is (0, 1).

[0029] A control system for laser wire feeding process parameters based on machine learning, the system has a program module corresponding to the above steps, and executes the steps in the above-mentioned control method for laser wire feeding process parameters based on machine learning when running.

[0030] A computer-readable storage medium stores a computer program, wherein the computer program is configured to implement the steps of a method for controlling laser wire feeding process parameters based on machine learning when called by a processor.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] The present invention utilizes process parameter data samples, and can better predict the height and width of deposit beads through process parameters, and reversely predict the required process parameters through the required bead width and bead height. Therefore, it can reduce the shape deviation of the deposit caused by the process personnel's incorrect judgment of the real-time observation of the molten pool, and can also realize the mining and analysis of equipment data, and give the equipment the ability to intelligently handle the process, thereby reducing the professional requirements of the process for the process personnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a flow chart of a method for controlling laser wire feeding process parameters based on machine learning in an embodiment of the present invention;

[0034] Figure 2 Schematic diagram of a machine learning algorithm in an embodiment of the present invention.

[0035] Description of reference numerals: DETAILED DESCRIPTION

[0036] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0037] Specific implementation plan 1: Combine Figure 1 and Figure 2 As shown, the present invention provides a method for controlling laser wire feeding process parameters based on machine learning, comprising the following steps:

[0038] S100, on the same test plate, respectively changing the process parameters of the equipment to obtain different single-track deposits, measuring the bead height and bead width of the single-track deposits to obtain an initial data set, wherein the process parameters include laser power, scanning speed and wire feeding speed;

[0039] S200, dividing the initial data set collected in step S100 into a training set and a test set according to a certain ratio, training the training set using a BP neural network to obtain a training model, and then using the test set to detect the degree of fit of the training model to obtain the best fitting model;

[0040] S300, batch importing process parameters into the best fitting model obtained in step S200, obtaining the corresponding bead height and bead width data of the sediment, acquiring the database, and calling the process parameters in the database for processing according to the required layered shape.

[0041] The training set is trained using a BP neural network. Since the process parameters including laser power, scanning speed, and wire feeding speed are independent and unrelated, these parameters can be normalized and used as the input parameter X of the BP neural network. At the same time, the sediment indicators: bead width and bead height can be approximated as independent and unrelated variables as the output Y of the neural network. Due to the inline relationship between bead width, bead height and the three process parameters, the parameters can be optimized through neural network training, thereby obtaining a learning model with a higher accuracy.

[0042] Given training samples S = {(X1, Y1), (X2, Y2), ..., (X n , Y n )},X i ∈R 3 , Y i ∈R 2 , X is the normalized process parameter data, R 3 is the dimension of the input data, namely, laser power x1, scanning speed x2, wire feeding speed x3. Y represents the sediment index, R 2 Is the dimension of the output data, namely, bead width y1 and bead height y2. Figure 2Describes a BP neural network with 3 input neurons, 2 output neurons, and a single hidden layer. If the threshold of the jth neuron in the output layer is set to σ j Indicates that the threshold of the mth neuron in the hidden layer is τ m Indicates that the connection weight between the i-th neuron in the input layer and the m-th neuron in the hidden layer is λ im , the connection weight between the mth neuron in the hidden layer and the jth neuron in the output layer is μ mj . Then the input received by the mth neuron in the hidden layer is:

[0043]

[0044] The input received by the jth neuron in the output layer is:

[0045]

[0046] Among them, h m represents the value of the mth neuron in the hidden layer, and q represents the total number of q neurons in the hidden layer;

[0047] If the discriminant function of the hidden layer and the output layer uses the Sigmoid function For the training set, assume that the current neural network output is Then the mean square error of the BP neural network on the data set is:

[0048]

[0049] In this model, a total of (3+2+1)q+2 parameters need to be determined, and the gradient descent algorithm is used to calculate the target E k The negative gradient direction is used to adjust the parameters. k , given the learning rate ξ, after one round of learning:

[0050] Δμ hj =ξf j h m

[0051] Δσ j =-ξf j

[0052] Δλ ih =ξl h x i

[0053] Δτ h =-ξl h

[0054] in: The learning rate ξ controls the step size of each round of learning iteration and takes a value of (0, 1).

[0055] The BP neural network was used to train and test the data in Table 1. The R2 score of the final model was 0.842 on the test set and 0.921 on the training set.

[0056] Table 1

[0057]

[0058]

[0059] Specific implementation scheme two: The present invention provides a control system for laser wire feeding process parameters based on machine learning. The system has a program module corresponding to the above steps, and executes the steps in the above-mentioned control method for laser wire feeding process parameters based on machine learning when running.

[0060] The other combinations and connection relationships of this embodiment are the same as those of the first embodiment.

[0061] Specific implementation scheme three: The present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program is configured to implement the steps of a method for controlling laser wire feeding process parameters based on machine learning when called by a processor.

[0062] The other combinations and connection relationships of this embodiment are the same as those of the first embodiment.

[0063] Although the present invention is disclosed as above, the protection scope of the present invention is not limited thereto. Those skilled in the art may make various changes and modifications without departing from the spirit and scope of the present invention, and these changes and modifications will fall within the protection scope of the present invention.

Claims

1. A method for controlling laser wire feeding process parameters based on machine learning, characterized in that: The following steps are involved: S100, on the same test plate, respectively changing the process parameters of the equipment to obtain different single-track deposits, measuring the bead height and bead width of the single-track deposits to obtain an initial data set, wherein the process parameters include laser power, scanning speed and wire feeding speed; S200, dividing the initial data set collected in step S100 into a training set and a test set according to a certain ratio, training the training set using a BP neural network to obtain a training model, and then using the test set to detect the degree of fit of the training model to obtain the best fitting model; S300, batch importing process parameters into the best fitting model obtained in step S200, obtaining the corresponding bead height and bead width data of the sediment, acquiring the database, and calling the process parameters in the database for processing according to the required layered shape.

2. The method for controlling laser wire feeding process parameters based on machine learning according to claim 1, characterized in that: In step S200, the laser power, scanning speed and wire feeding speed are set as input parameters X of the BP neural network, and the bead width and bead height are set as output Y of the neural network; Given training samples S = {(X1, Y1), (X2, Y2), ..., (X n , Y n )},X i ∈R 3 , Y i ∈R 2 , X is the normalized process parameter data, R 3 is the dimension of the input data, i.e., laser power x1, scanning speed x2, wire feeding speed x3; Set the threshold of the jth neuron in the output layer with σ j Indicates that the threshold of the mth neuron in the hidden layer is τ m Indicates that the connection weight between the i-th neuron in the input layer and the m-th neuron in the hidden layer is λ im , the connection weight between the mth neuron in the hidden layer and the jth neuron in the output layer is μ mj ; The input received by the mth neuron in the hidden layer is: The input received by the jth neuron in the output layer is: Among them, h m represents the value of the mth neuron in the hidden layer, and q represents the total number of q neurons in the hidden layer.

3. The method for controlling laser wire feeding process parameters based on machine learning according to claim 2, characterized in that: The discriminant function of the hidden layer and the output layer uses the Sigmoid function For the training set, assume that the current neural network output is Then the mean square error of the BP neural network on the data set is:

4. The method for controlling laser wire feeding process parameters based on machine learning according to claim 3 is characterized in that: Using the gradient descent algorithm, the target E k The negative gradient direction is used to adjust the parameters. k , given a learning rate ξ, after one round of learning: Dm hj =ξf j h m Board j =-ξf j Dl ih =ξl h x i Dt h =-ξl h in: The learning rate ξ controls the step size of each learning iteration.

5. The method for controlling laser wire feeding process parameters based on machine learning according to claim 4, characterized in that: The value of ξ is (0, 1).

6. A control system for laser wire feeding process parameters based on machine learning, characterized in that: The system has a program module corresponding to the steps of any one of claims 1 to 5 above, and executes the steps in the above-mentioned method for controlling laser wire feeding process parameters based on machine learning when running.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is configured to implement the steps of the method for controlling laser wire feeding process parameters based on machine learning described in any one of claims 1 to 5 when called by a processor.

Images